079 - One Developer Used AI to Build Adobe Apps. Now He's Taking on Microsoft Office.
Learn AI in BitsShow Notes
After drawing attention earlier in October for using AI to build open-source alternatives to seven Adobe applications in Rust, developer Brandon Thomas and the ArtCraft project have moved into Microsoft Office territory with three new projects: WordCraft, GridCraft, and DeckCraft.
WordCraft targets Word, with a familiar ribbon, styles, tables, track changes, references, and mail merge, reading and writing Word documents and exporting to PDF, OpenDocument, rich text, Markdown, and plain text. GridCraft targets Excel, supporting spreadsheet formulas, dynamic arrays, tables, charts, conditional formatting, and PivotTables, opening and saving Excel workbooks as a native app or in a browser. DeckCraft targets PowerPoint, supporting slide decks with themes, layouts, shapes, images, tables, charts, transitions, and animations, with the ability to open and save PowerPoint files. All three are written in Rust, open source under permissive licenses, and described by their own repositories as clean-room reimplementations built from scratch rather than from Microsoft's proprietary code.
Thomas says he used Anthropic's Claude Opus 5.5 to help build the earlier Adobe-style apps, and the Office-style projects follow the same approach: AI coding agents work across the codebase, implement features, write tests, and fix problems while Thomas directs the work and checks the results. Each app also exposes its actions as commands through a command-line interface and an MCP, or Model Context Protocol, server, letting an AI agent ask the app to do something, such as generating a spreadsheet or formatting a document, and then inspect the result.
The episode is direct about where things stand: all three remain in alpha, and their own public roadmaps describe clear gaps between covering a feature by name and matching Microsoft Office in professional work, since a spreadsheet can implement hundreds of functions and still behave differently on a complex financial model. Building a clone doesn't mean the original has been replaced; feature coverage, compatibility, and stability are what will decide that. The episode's bigger point is about the shifting economics of rebuilding mature software: what used to require a large engineering team and years of work can now be attempted by a small team directing AI agents, even as the hardest remaining work moves toward architecture, testing, and judging whether the result is dependable enough to use.
Sources & References Thurrott: Open Source AI Dev is Going After Microsoft Office, Too — https://www.thurrott.com/office/342646/open-source-ai-dev-is-going-after-microsoft-office-too WordCraft: GitHub repository, clean-room reimplementation of Microsoft Word in Rust — https://github.com/storytold/wordcraft GridCraft: GitHub repository, clean-room spreadsheet/Excel-style app in Rust — https://github.com/storytold/gridcraft DeckCraft: GitHub repository, clean-room reimplementation of Microsoft PowerPoint in Rust — https://github.com/storytold/deckcraft Flowtivity: Someone Rebuilt Adobe in Rust with AI Agents. Is It Usable? — https://flowtivity.ai/blog/adobe-rust-cleanroom-clones/ Ars Technica: "Software is over": Bold AI developer takes aim at Adobe with open source clones — https://arstechnica.com/ai/2026/10/software-is-over-bold-ai-developer-takes-aim-at-adobe-with-open-source-clones/
Voice narration is AI-generated.
More from Learn AI in Bits
A wave of cybersecurity disclosures has made this week feel like a turning point, with people calling it "Black October." This episode treats it as a cluster of high-impact incidents, not proof that one AI-driven campaign caused them all, and separates where AI was directly involved from where the failures were conventional.
Recursive self-improvement is the idea that an AI can improve the very process it uses to improve itself, and this episode separates what's been demonstrated so far from the dramatic predictions often attached to the term. It draws a clear line between self-correction (revising one answer), automated optimization (keeping the best of several tested variants), and true recursive improvement, where the system changes the process that generates future improvements, so the next round starts with a better improver.
This week's AI News Sunday Wrap-Up covers a week where the industry moved on several fronts at once: new flagship and open models, enterprise agent governance, a serious safety warning, legal pressure, and the physical infrastructure limits behind the AI race.
Meta's Muse story is moving fast, and this episode takes a closer look at Muse Spark, the multimodal reasoning model family behind Meta's push into AI agents, including yesterday's release of an open-weight version for developers.
When a large language model generates text one token at a time, even a fast GPU hits a sequential bottleneck. This episode explains speculative decoding, an inference technique that prepares several tokens at once and checks them together, and when the extra computation pays off.